Aleksandra Fostikov, Boban Petrovski, Challenges of Using Generative Artificial Intelligence (GAI) in the “Research” of Personas from National History

Abstract of the publication

Recent breakthroughs in the development of Generative Artificial Intelligence (GAI) have created opportunities to test it in nearly all conceivable fields and sciences. While the positive effects of their usage are often highlighted, notable concerns have emerged, particularly regarding the potential for misinformation, which could create significant problems, especially in the realm of historical facts and contexts. The issue of GAI’s validity, given that it is not trained as a historian with an analytic skill set, emerged as the most relevant one in our research. In evaluating one GAI`s capability to distinguish between misinformation and valid data, we simultaneously tested in real-time three of the most popular linguistic chatbot models with the same set of questions in more than one language about two relevant personas from Medieval History. These personas were chosen because of their notable importance to three modern Balkan nations and the inherent challenges they present for GAI analysis.